On Clustering Bodies: Geometry and Polyhedral Approximation

Abstract

The present paper studies certain classes of closed convex sets in finite-dimensional real spaces that are motivated by their application to convex maximization problems, most notably, those evolving from geometric clustering. While these optimization problems are ℕℙ-hard in general, polynomial-time approximation algorithms can be devised whenever appropriate polyhedral approximations of their related clustering bodies are available. Here we give various structural results that lead to tight approximations.